Characterizing and Verifying Queries Via CINSGEN

Hanze Meng, Zhengjie Miao, Amir Gilad, Sudeepa Roy, Jun Yang

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review


Example database instances can be very helpful in understanding complex queries. Different examples may illustrate alternative situations in which answers emerge in the query results and can be useful for testing. Examples can also help reveal semantic differences between queries that are supposed to be equivalent, e.g., when students try to understand how their queries behave differently from a reference solution, or when programmers try to pinpoint mistakes inadvertently introduced by rewrites meant to improve readability or performance. In this paper, we propose to demonstrate CinsGen, a system that can characterize queries and help distinguish between two queries. Given a query, CinsGen generates minimal conditional instances (c-instances) that satisfy it. In turn, each c-instance is a generalization of multiple database instances, yielding a compact representation. Thus, using CinsGen enables users to obtain a comprehensive and compact view of all scenarios that satisfy a specified query, allowing for query characterization or distinction between two queries.

Original languageEnglish
Title of host publicationSIGMOD 2023 - Companion of the 2023 ACM/SIGMOD International Conference on Management of Data
PublisherAssociation for Computing Machinery
Number of pages4
ISBN (Electronic)9781450395076
StatePublished - 4 Jun 2023
Externally publishedYes
Event2023 ACM/SIGMOD International Conference on Management of Data, SIGMOD 2023 - Seattle, United States
Duration: 18 Jun 202323 Jun 2023

Publication series

NameProceedings of the ACM SIGMOD International Conference on Management of Data
ISSN (Print)0730-8078


Conference2023 ACM/SIGMOD International Conference on Management of Data, SIGMOD 2023
Country/TerritoryUnited States


  • database usability
  • incomplete database


Dive into the research topics of 'Characterizing and Verifying Queries Via CINSGEN'. Together they form a unique fingerprint.

Cite this